Skip to main navigation Skip to search Skip to main content

Power reduction technique in coefficient multiplications through multiplier characterization

  • Stony Brook University
  • IBM

Research output: Contribution to journalArticlepeer-review

Abstract

This paper presents a multiplier power reduction technique for low-power DSP applications through utilization of coefficient optimization. The optimization is implementation dependent in that the multipliers are assumed to be designed in either ASIC or full-custom architectures for general purpose multiplication. The paper first describes a model characterizing the power consumption of the multiplier. Then the coefficient optimized made based on this model. This methodology is applicable to multiplications requiring a large set of coefficients and random data sets. We can accurately estimate the actual power dissipation of the multipliers using the characterization technique. The coefficient optimization based on the power model can save as much as 34.02%.

Original languageEnglish
Pages (from-to)101-113
Number of pages13
JournalJournal of VLSI Signal Processing Systems for Signal, Image, and Video Technology
Volume38
Issue number2
DOIs
StatePublished - Sep 2004

Keywords

  • Coefficient optimization
  • Low-power multiplier
  • Power modeling
  • Power weight factor

Fingerprint

Dive into the research topics of 'Power reduction technique in coefficient multiplications through multiplier characterization'. Together they form a unique fingerprint.

Cite this